Practical overview
When it fits
Consider it when you need a deployable codebase for experimenting with medical assistance or research workflows that combine document RAG, web search and image analysis, and can configure the required service credentials.
Who it is for
- Healthcare professionals seeking a configurable medical-assistance chatbot project.
- Researchers working with medical documents and research retrieval.
- Patients interested in the project’s stated patient-interaction use case.
- Developers studying multi-agent workflows and retrieval-augmented generation.
Documented uses
- Prepare medical PDFs for retrieval by extracting their text, tables and images.
- Retrieve relevant medical-document passages using hybrid search and rerank the results.
- Support medical research with document references and web-search retrieval.
- Explore medical-image analysis workflows for chest X-rays and skin lesions.
Requirements & limits
The repository is public and lists an Apache-2.0 license. The documented deployment requires Docker or manual installation and API credentials for configured services.
- The README documents Python 3.11+ and version badges for LangGraph, LangChain, Qdrant, FastAPI, Pydantic and Docling.
- The RAG pipeline uses Docling parsing, Qdrant hybrid search and cross-encoder reranking.
- Azure OpenAI is the documented development configuration; alternate model providers may need code changes.
- Speech, web search and reranking configuration uses ElevenLabs, Tavily and a Hugging Face token, respectively.
- Chest X-ray classification and skin-lesion segmentation are listed; brain-tumor detection is identified as a future integration.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Prepare the project for Docker deployment
Build and start the application container with the documented environment configuration.
Before you start
- Docker installed
- API keys for the required services
- Repository files and a configured .env file
Steps
Clone the repository and change into its directory.
Create a .env file with the required model, embedding, speech, web-search and reranker credentials; configure Qdrant credentials only when using a Qdrant server.
Build the image with docker build -t medical-assistant .
Run the container with docker run -d --name medical-assistant-app -p 8000:8000 --env-file .env medical-assistant.
Expected outcome
The README says the application is available at http://localhost:8000.
Guide limitations
- The documented development configuration uses Azure OpenAI; changing providers may require code modifications.
- Using local models may require changes throughout the codebase, especially in the agents.
- The documented run step depends on correctly configured credentials and provider permissions.
Ingest medical documents into the vector database
Add one PDF or a directory of documents to the retrieval data using the documented Docker commands.
Before you start
- The application container started as medical-assistant-app
- Documents available at paths accessible inside the container
Steps
For one document, run docker exec medical-assistant-app python ingest_rag_data.py --file ./data/raw/brain_tumors_ucni.pdf.
For multiple documents, run docker exec medical-assistant-app python ingest_rag_data.py --dir ./data/raw.
Expected outcome
The selected document or directory is passed to the project's RAG ingestion script for vector-database ingestion.
Guide limitations
- The README documents these commands for the Docker setup.
- The provided single-file example uses a specific path that must be available inside the container.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 17 October 2026
- Original website title
- souvikmajumder26/Multi-Agent-Medical-Assistant · Official homepage title · retained in its source language